Introduction

The ESKAPE group, comprising Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter sp. are regarded as the most dangerous MDR, extensively drug-resistant (XDR), and pandrug-resistant (PDR) strains currently known1. Among these, multidrug-resistant (MDR) A. baumannii causes a variety of severe, life-threatening infections, especially in intensive care unit (ICU) patients, including ventilator-associated pneumonia, bloodstream infections, urinary tract infections, and surgical site infections2.

Disturbingly, the global incidence of MDR A. baumannii clinical isolates has increased by 40% over the previous decade3. The European Centre for Disease Prevention and Control (ECDC) indicates that the share of Acinetobacter spp. isolates from ICUs in Poland increased from 60% of all ICU samples in 2017 to 69% in 20214. In 2021, resistance to carbapenems in A. baumannii in Poland reached 82.7%, more than doubling the EU population-weighted mean of 39.9%. In Polish ICUs, acquired multidrug-resistant (MDR) A. baumannii strains, particularly those resistant to carbapenems, have become increasingly predominant, accounting for over 50% of A. baumannii infections reported to the European Antimicrobial Resistance Surveillance Network (EARS Net) in recent years5. The prevalence of A. baumannii was heavily influenced by the COVID-19 pandemic, with 55.6% of COVID-19 patients in ICUs being coinfected with CRAB (carbapenem-resistant A. baumannii) isolates. Among all MDR isolates identified, A. baumannii alone accounted for 52% of infections by the ESKAPE group6.

As with other bacterial infections, those caused by A. baumannii can be monitored using genotyping: the results can be used to effectively identify sources of infection and links in the chain of transmission. The resistome of A. baumannii is highly diverse and includes mechanisms such as β-lactamase production, efflux pumps, membrane protein alterations, ribosomal methylation and plasmid-borne resistance determinants, which together contribute to the rapid emergence of MDR, XDR and even PDR phenotypes7. The species possesses an open pan-genome with high plasticity, enriched by insertion sequences, transposons, integrons and prophages, enabling continuous acquisition of novel resistance genes and expansion of its genomic pool7. Importantly, resistance islands and mobile genetic elements act as major hotspots of antimicrobial resistance determinants, including carbapenem-hydrolyzing blaOXA variants7. This genomic versatility, combined with a high rate of horizontal gene transfer, forms the basis of the global health threat posed by A. baumannii8.

The β-lactamases produced by A. baumannii are divided into four molecular classes (A, B, C, and D), with class B metallo-β-lactamases (MBLs) and class D OXA-type carbapenemases being of greatest clinical significance due to their ability to hydrolyze carbapenems9. Given the urgent need to investigate the spread of nosocomial strains in Poland, the aim of this study was to determine the genetic profile of A. baumannii isolates in Polish hospitals with the use of Pulsed Field Gel Electrophoresis (PFGE) and Multilocus Sequence Typing (MLST).

Methods

Bacterial isolates

A total of 38 clinical isolates of A. baumannii were collected retrospectively from the strain collections of three tertiary hospitals in Warsaw and Wrocław, Poland. All samples were collected between 13.02.2008 and 30.08.2019. Most studied patients were male (78.9%) aged 20–80 years; only eight (21.1%) were female, aged 55–88 years. The bacterial samples were isolated following a request for microbiological examination. Consent for diagnostic and therapeutic procedures was obtained from the patient or legal guardian at the time of hospital admission, as part of the standard medical care process. In some cases, samples were obtained for bacteriological analysis by bronchoscopy; as this is a high-risk procedure, patients gave their consent for it to be done, after being made aware of the associated risks. All methods were performed in accordance with the Declaration of Helsinki.

Identification of nearly all A. baumannii clinical isolates was performed by MALDI-TOF MS (score > 2.000); of the remainder, isolate K6 was identified using the GNI test (VITEK system, bioMérieux, France), and isolates K4 and K10 were identified using the ID 32E test (bioMérieux). Origin from invasive infection sites, such as blood, bronchial aspirates, and respiratory secretions, is indicated in Table 1. Thus, the collection focused on A. baumannii strains associated with clinically-significant infections, primarily bloodstream and respiratory tract infections.

Additionally, five A. baumannii isolates from pig manure were analysed. These were obtained in 2019, with the use of the selective CHROMagar Acinetobacter medium (CHROMagar, France). Pure cultures were obtained through three consecutive streakings and incubated under manufacturer-recommended conditions. Identification was performed by MALDI-TOF MS/MS (score > 2.000) at ALAB Laboratoria sp. z o.o. (Warsaw, Poland). The bacteria were stored as glycerol stocks (40% v/v) at − 80 °C in LB broth medium (Sigma, USA). The origin and characteristics of all 45 isolates used in this study are presented in Table 1.

Table 1 The origin and characteristics of the isolates and reference strains of A. baumannii.
Full size table

Drug susceptibility test

The drug susceptibility of the bacteria in the clinical isolates was tested using the Kirby-Bauer disk diffusion method with the following disks (BioMaxima, Lublin, Poland): cefepime (30 µg), ceftazidime (10 µg), imipenem (10 µg), meropenem (10 µg), amikacin (30 µg), gentamicin (10 µg), ciprofloxacin (5 µg), trimethoprim/sulfamethoxazole (1.25–23.75 µg); the E-test for ampicillin/sulbactam, imipenem, meropenem, gentamicin, amikacin, colistin was used to determine MIC (minimal inhibitory concentration) (0.016 to 256 µg/ml). The colistin susceptibility of isolates K5, K9, K11 and K17 was determined based on MIC using the microdilution method (0.0625 to 64 µg/ml) (MICRONAUT MIC-Strip Colistin, Bruker, Germany).

The tested antibiotics belong to the following classes: aminoglycosides (amikacin, gentamicin), β-lactams including penicillins and carbapenems (ampicillin/sulbactam, imipenem, meropenem), cephalosporins (cefepime, ceftazidime), fluoroquinolones (ciprofloxacin), sulfonamides (trimethoprim/sulfamethoxazole), and polymyxins (colistin). Antimicrobial susceptibility was interpreted based on the applicable guidelines at the time: CLSI in 2008–200912,13, and the regularly-updated EUCAST recommendations (version 1–3 in 2010–2013 and version 4–7 in 2014–2017)14.

The total number of isolates (N) tested differed for each drug due to the individually-selected drug susceptibility profiles for each isolate. Increased exposure to antimicrobial agents was inferred from the range and concentrations of antimicrobials applied during susceptibility testing, and isolates showing such characteristics were classified as resistant (R).

The drug susceptibility of the isolates from pig manure was determined using the automated VITEK® 2 Compact system (bioMérieux), in accordance with EUCAST guidelines. AST-N331 cards were used for Acinetobacter spp., providing accurate phenotypic susceptibility profiles. It should be emphasized that the clinical and pig manure isolates were examined using distinct antimicrobial susceptibility testing (AST) methods. In addition, although all tests followed EUCAST guidelines, the sensitivity and precision of resistance detection may be influenced by methodological differences, particularly for certain classes of antibiotics. Definitions of MDR and XDR were adopted from the international expert consensus proposed by Magiorakos et al. According to this classification, MDR was defined as non-susceptibility to at least one agent in three or more antimicrobial classes, while XDR was defined as non-susceptibility to at least one agent in all but two or fewer antimicrobial classes15.

Identification of β-lactamase-encoding genes

Genomic DNA from A. baumannii isolates was extracted using the Bacterial & Yeast Genomic DNA Purification Kit (Eurx, Poland), according to the manufacturer’s instructions. The extracted DNA was stored at − 20 °C for subsequent analyses. A set of genes encoding common β-lactamases was amplified, including class D β-lactamase genes (blaOXA−2316, blaOXA−51, blaOXA−5817, and blaOXA−4016; class A β-lactamase genes (blaTEM, blaGES18; and class B β-lactamase genes (blaIMI, blaIMP19, and blaNDM16. Multiplex polymerase chain reaction (PCR) assays were performed to amplify the following carbapenemase genes: blaOXA−40 (402 bp), blaOXA−23 (718 bp), and blaNDM (517 bp) in one reaction; blaIMI (206 bp) and blaIMP (597 bp) in a second multiplex set; and blaOXA−58 (599 bp) together with blaOXA−51 (353 bp) in a third multiplex group. Additionally, blaTEM (1080 bp) and blaGES (370 bp) were amplified in separate, individual PCR reactions.

PCR amplification was carried out using Platinum SuperFi II DNA Polymerase (Invitrogen, MA, USA). Multiplex PCR was performed under the following thermal cycling conditions: initial denaturation at 98 °C for 30 s; 30 cycles consisting of denaturation at 98 °C for 10 s, annealing at 60 °C for 30 s, and extension at 72 °C for 43 s; followed by a final extension at 72 °C for 5 min. For the individual amplification of blaTEM and blaGES, the protocol included an initial denaturation at 98 °C for 30 s, 35 cycles of denaturation at 98 °C for 10 s, annealing at 60 °C for 10 s, and extension at 72 °C for 30 s, with a final extension step of 5 min at 72 °C.

The PCR products were analysed using gel electrophoresis. To verify the specificity of primer binding and to ensure that no non-specific amplification of β-lactamases genes occurred, strains isolated from pig manure, confirmed to be β-lactamase-susceptible, were used as negative controls. Primer data and detailed information on detected genes are described in Table 2.

Table 2 Primers employed for the detection of β-lactamase genes in A. baumannii.
Full size table

MLST

A single pure colony of A. baumannii clinical isolate was inoculated into LB broth (Sigma) and cultured aerobically at 37 °C with vigorous shaking (120 rpm) overnight. The genomic DNA was extracted from the overnight cultures using the Genomic Mini kit (A&A Biotechnology, Poland), according to the manufacturer’s instructions. Seven housekeeping genes (gltA, gyrB, gdhB, recA, cpn60, gpi, rpoD) were amplified, the conditions and primers are given in the common Oxford MLST scheme on the MLST Database website20 (Table 3). However, the analysis of the recA gene required a newly-designed forward (Fw) primer, due to the formation of non-specific amplification products observed with the original primer (Table 3); the PCR products were cleaned using the Short DNA Clean-Up-DNA Purification Kit (EURx, Poland).

The housekeeping genes were sequenced by Genomed SA (Poland) and Eurofins (Germany). The sequencing results were analysed using Chromas 2.6.6 (Technelysium Pty Ltd, Australia) and CloneManager Professional Suite 8.0 (Sci-Ed, CO, USA) software. The locus number of each gene was assigned by comparison with the MLST database. Any new alleles or STs identified were submitted to the MLST database20. Furthermore, MLST analysis was also performed for the reference strains A. baumannii ATCC 19606 and ATCC 17978, utilizing both the NCBI and PubMLST databases. Sequence types (STs), of all A. baumannii isolates were analysed using Phyloviz 2.0, available at http://www.phyloviz.net/21. Isolates sharing six out of seven alleles (SLVs, single-locus variants) were considered clonal. All STs discovered in the study were subjected to minimum spanning tree (MST) analysis to determine their relevance in MLST.

Table 3 Housekeeping genes used in MLST analysis and primer sequences.
Full size table

PFGE

PFGE was performed as described previously22. Briefly, an overnight bacterial culture with an OD600 of 0.7 was diluted and mixed with a cell suspension buffer (100 mM Tris, 100 mM EDTA, pH 8.0) containing Proteinase K (EURx). An equivalent volume of 1% low melting agarose (BioRad, CA, USA) with 1% SDS was added, and the resulting mixture was distributed into a plug mold. The agarose plugs containing genomic DNA were lysed in a cell lysis solution with Proteinase K, washed, and digested with ApaI restriction enzyme (EURx). DNA size was estimated using the Lambda PFGE Ladder (New England Biolabs, MA, USA). Electrophoresis was performed in 1% low melting agarose (BioRad) gels using a pulsed-field electrophoresis system (Chef Mapper XA, BioRad) under the following conditions: temperature 14 °C, voltage 6 V/cm, and switch ramp of 5–20 s for 19 h. The gels were then stained with ethidium bromide solution (1 µg/mL) for 30 min and de-stained for 45 min in distilled water with two water changes.

PFGE cluster analysis

PFGE profiles (PFGE electropherograms) were analysed using BioNumerics software ver. 7.6.2 (Applied Maths, bioMérieux Company, France). A dendrogram illustrating genetic relatedness was generated using the unweighted pair group method with mathematical averaging (UPGMA). The DNA similarities were calculated by applying the band-based Dice coefficient23 with the following settings: band tolerance 1.5%, optimization 1.5% and tolerance change 1%. Uncertain bands were ignored. Pulsotypes were considered to represent the same cluster if the DNA pattern homology was > 80%.

Fig. 1
figure 1

A schematic of the methodology.

Full size image

Figure 1.Schematic of the methodology. Created in https://BioRender.com.

Results

Susceptibility to antimicrobial agents

The results of the drug susceptibility tests are presented in Tables 4 and 5. Briefly, 97.3% of A. baumannii clinical isolates tested were found to be XDR, and only 2.7% were MDR (Suppl. Table 1). In addition, all tested isolates were found to be resistant to cefepime, ceftazidime, imipenem, meropenem and ciprofloxacin (Table 4). All but one (97.3%) of the clinical isolates were found to be susceptible to colistin (Table 4).

All isolates from pig manure were found to be susceptible to cefepime, imipenem, meropenem, gentamicin, amikacin, and trimethoprim/sulfamethoxazole (Table 5). Furthermore, all were resistant to ciprofloxacin, and only one (Ś2) showed resistance to ceftazidime (Table 5).

Table 4 The results of drug susceptibility testing of A. baumannii clinical isolates.
Full size table
Table 5 The results of drug susceptibility testing of A. baumannii isolates from pig manure.
Full size table

Statistical methods

The differences in MIC results between the two groups of strains (i.e. clinical and pig manure) were assessed using Fisher’s exact test. The test showed significant differences in antibiotic resistance between the groups (p < 0.05).

Identification of β-lactamases-encoding genes

The genetic determinants responsible for β-lactam antibiotic resistance, particularly carbapenems, were determined by PCR amplification using specific primers targeting nine known β-lactamase-encoding genes. In total, the analysis included 43 A. baumannii isolates, comprising carbapenem-resistant clinical isolates and carbapenem-susceptible isolates from pig manure, the latter serving as negative controls for carbapenemase gene detection. The PCR screening revealed the presence of several genes from the OXA-type β-lactamase family (Table 6). The blaOXA−51 gene, considered an intrinsic marker of A. baumannii, was detected in 42 of all 43 isolates (98%), including all those from pig manure, and 37 out of 38 clinical isolates (97%), thus confirming the presence of the species. The acquired carbapenemase gene blaOXA−23 was identified in 13 clinical isolates (13/38 clinical isolates, 43%), indicating a significant contribution of this gene to the observed carbapenem resistance in the tested population.

Other OXA-type genes associated with carbapenemase activity were detected at lower frequencies: blaOXA−40 was present in nine out of 38 clinical isolates (24%), and blaOXA−58 in six out of 38 clinical isolates (16%). Among the extended-spectrum β-lactamases (ESBL), blaTEM were detected in nine of the 38 clinical isolates (24%) (Table 6; Suppl. Table 2; Suppl. Figure 1). These results indicate that OXA-type carbapenemases, particularly OXA-23, are the predominant β-lactamases associated with carbapenem resistance in this set of A. baumannii isolates, with no evidence of metallo-β-lactamases (e.g., NDM, IMP), or class A carbapenemases (e.g., GES, IMI) in this population.

Table 6 The prevalence of β-lactamase genes among 38 carbapenem-resistant A. baumannii clinical isolates from three hospitals in Poland between 2008 and 2019.
Full size table

Genotyping by MLST analysis

Among 38 A. baumannii isolates, nine STs were identified in accordance with the Oxford MLST Scheme (Fig. 2; Suppl. Table 3). Three new STs were noted: ST3281 (n = 1, 2.6%, K3 isolate) and ST3282 (n = 1, 2.6%, K6 isolate), both characterised by allele combinations unreported in the MLST database, and ST3285 (n = 2, 5.3%, K18 and K20 isolates) characterised by a new allele variant of the gdhB gene denoted as 345. All new STs were isolated in the same hospital. One clonal complex comprised ST195 (n = 13, 34.2%), together with three (single-locus variants, SLVs), viz. ST208 (n = 6, 15.8%), ST425 (n = 11, 28.9%), and ST3285 (n = 1, 2.6%); this clonal complex also included ST218 (n = 1, 2.6%) and its SLV, ST348 (n = 2, 5.3%) (Fig. 2). Another clonal complex comprised ST440 (n = 1, 2.6%) and its SLV, ST3282. The newly-discovered ST3281 was a singleton and did not belong to any cluster. The goeBURST analysis showed ST195 to be a founder of the largest clonal complex (Fig. 2b). Furthermore, MLST analysis identified sequence type ST931 in A. baumannii reference strain ATCC 19606 and ST112 in ATCC 17978.

Fig. 2
figure 2

(A). goeBURST distance analysis indicating clonal complexes of A. baumannii isolates. The numbers inside the squares indicate the ST types. The size of the square indicates the number of isolates, with the largest size associated with the largest number of isolates. Light green indicates group founders, and light blue for common nodes. (B) goeBURST Full MST; type affinity with the founder ST195 (corresponding to the yellow node). The numbers next to the lines indicate the locus variants between the STs, and correspond to different grayscale colours (darker links have fewer mismatches between the STs than lighter grey links).

Full size image

Molecular genotyping by PFGE

PFGE profiling of 40 A. baumannii isolates (i.e. 38 clinical isolates and two reference strains), produced seven major clusters (A-G), each comprising three or more isolates (Fig. 3). Cluster A contained 14 clinical isolates from three different hospitals. Isolates AB9 and AB10 sharing the same ST, were collected from the same hospital and had identical pulsotypes. Cluster A also included the genetically-similar isolates o2251 and o2252 from Warsaw hospital; these had the same ST, together with isolate 1724 originated from hospital in Wroclaw (Fig. 3). Among four isolates that exhibited new STs, three (K20, K6, K18) were placed in three different clusters, while the fourth (K3) was a singleton (Fig. 3). Additionally, incorporating five isolates from pig manure into the analysis, the PFGE results revealed that three of these isolates shared a genotype closely related to ATCC 17978, clustering together in cluster C. The remaining two isolates from pig manure were assigned to cluster A, showing genetic similarity to certain clinical isolates (Fig. 3; Suppl. Figure 2).

Fig. 3
figure 3

Dendrogram of the PFGE profiles of A. baumannii isolates with regard to MLST analysis (STs). Seven clusters (A-G), each marked with different coloured lines and squares, were identified using a cut-off value above 80% for the similarity coefficient. Analysis was performed by bionumerics software.

Full size image

The results are summarized in a schematic overview in Figure. 4

Fig. 4
figure 4

Schematic overview summarizing the main results of the study. Created in https://BioRender.com.

Full size image

Discussion

Due to its wide repertoire of antibiotic resistance mechanisms, A. baumannii plays a growing role in healthcare-associated infections24,25. As such, there is an urgent need for ongoing epidemiological studies involving genotyping aimed at preventing outbreaks of infection in the ICUs in Poland and across Europe.

The aim of the study was to determine the genetic structure of A. baumannii isolates in Polish hospitals. These structures were identified using MLST and PFGE, with MLST tracking global clonal structures26 and PFGE used to define infection outbreaks27.

Our analysis identified nine STs among the 38 isolates, three of which were newly discovered (ST3281, ST3282, ST3285). Our findings also indicate a new allele of the gdhB gene (345), which was detected in two isolates from a hospital in Warsaw. All three novel STs (ST3281, ST3282, ST3285) were isolated from patients in the same tertiary hospital and exhibited XDR profiles. Although these sequence types have not been previously reported in public MLST databases, their detection in our clinical cohort does not yet indicate clonal expansion or dissemination. Therefore, they are herein referred to as previously undescribed, rather than emerging, STs. The most abundant STs in our study (ST195, ST425 and ST208 and ST348) have previously been reported in Poland28. These STs belong to clonal complex 92 (CC92/IC2), which is widespread globally, as well as in Poland29,30,31,32,33,34; this complex is synonymous with ST2 according to Pasteur’s MLST scheme35. The singleton ST440, identified in our study, was classified as belonging to IC7, correlated with CC110, predominant in Central and South America36, with its novel SLV (ST3282). Notably, our newly-discovered ST3285 corresponds to the SLV of ST195, while ST3281 showed significant genetic distance from ST195. The appearance of these novel STs highlights the clonal diversity of A. baumannii and emphasizes the need for continued monitoring in clinical settings.

By December 2024, the Oxford database recorded 2538 A. baumannii isolates in Europe. This number includes 248 isolates found in Poland. In the present study, the largest group of isolates (n = 13), detected in three Warsaw hospitals, belong to ST195, reported across Europe. The second most common group of isolates (n = 11), identified in hospitals in Warsaw and Wroclaw, belong to ST425; this group has previously been detected in Poland, Greece, Italy, and Spain. Another group of isolated were associated with ST208 (n = 6), identified in two Warsaw hospitals; again, this group has been found to be widespread in sixteen European countries. Two isolates with ST348 were isolated from the same Warsaw hospital; these have been reported in Germany, Italy, Poland and Sweden20. A new finding in Poland concerned the presence of ST218 and ST440, which were identified in the same Warsaw hospital. Notably, a candidate for novel IC10 (CC33) has already been identified in Europe37.

PFGE produced ten distinct patterns, separated into seven clusters (A-G), and three unique patterns. The most abundant were clusters A and B comprising 34.8% and 21.7% of the isolates, respectively. Not all isolates within the same ST exhibited identical PFGE patterns, as noted previously38. As expected, none of the clinical or environmental isolates showed significant genetic similarity to reference strains ATCC 19606 or 17978, which are known to represent distinct, well-characterised laboratory lineages. AB9 and AB10 shared the same PFGE genotypes and ST (i.e. ST425), potentially indicating localised hospital transmission. It is noteworthy that all the newly-discovered STs were classified into separate clusters. Furthermore, the clinical isolates from three different hospitals were clustered within clone A, suggesting a potential scenario where isolates are transferred between ICUs within the same hospital or even between different hospitals.

Intriguingly, among the isolates tested for colistin resistance, only isolate o2252, belonging to the same ST (ST195) and clonal complex as o2251 (i.e. complex A), was resistant to colistin. Among the isolates tested in the present study, isolate K8 demonstrated a more sensitive antimicrobial profile than other members of ST208 classed as clone B. MLST typing found the widely-used reference strain ATCC 19606 to belong to ST931, and ATCC 17978 to ST112. These findings highlight the genetic divergence between reference strains and contemporary clinical isolates, further supporting the notion that historical laboratory strains may no longer reflect the current population structure of A. baumannii.

Interestingly, PFGE genotyping found that three A. baumannii isolates from pig manure (Ś2, Ś4, Ś6), were clustered together with the reference strain ATCC 17978, which was genetically distant from the clinical isolates. Significantly, the other two isolates (Ś1, Ś5) were assigned to the largest cluster (Cluster A), along with clinical isolates. Importantly, those isolates were obtained from the manure environment. The detection of common genotypic clusters between clinical isolates and from pig manure suggests possible gene flow and transmission between different ecological niches. This indicates a potential risk of A. baumannii strains, particularly those resistant to antibiotics, spreading from the hospital environment to the external environment and vice versa. This phenomenon is particularly concerning as these bacteria are capable of surviving advanced hospital wastewater treatment processes, which could lead to their further dissemination in the natural environment39. These results highlight the need to monitor sources and vectors of A. baumannii transmission and to strengthen preventive measures40.

The importance of this current report is underlined by the paucity of genotyping studies that have been performed on A. baumannii strains in Poland in the last 17 years, i.e. since 200728,41,42,43,44,45,46,47.

A report by EARS Net found that more than a half of Acinetobacter isolates in 25 European countries show resistance to at least one antimicrobial group, with the greatest prevalence of resistant isolates observed in Southern and Eastern Europe, and particularly high levels noted in Poland5. A major concern is the development of resistance to colistin, the last-line antibiotic treatment. Our present findings indicate an even higher percentage of resistant clinical isolates, with more than 97% of isolates classified as XDR, and over 2% as MDR. A study of A. baumannii isolates from Polish hospitals in 2013 found 60% to be resistant to all tested antimicrobials except colistin, and 81% as XDR48. Similarly, Kasperski et al. reported that 37% of A. baumannii isolates were resistant to all tested antibiotics except colistin, with 77% classified as XDR16. A recent study of COVID-19 patients in Cracow found 81.5% of Acinetobacter spp. isolates to be XDR49.

The predominance of blaOXA−23 among acquired carbapenemases in our study (43%) is consistent with Kasperski et al.16, who found the gene to be present in 44% of bloodstream infection strains in Poland. However, significant differences between studies were observed regarding the prevalence of blaOXA−40 and blaOXA−58: Kasperski et al. report a higher prevalence of blaOXA−40 in clinical isolates, i.e. 42%, compared to our 24%, but a lower prevalence of blaOXA−58I, i.e. absent, compared to 16% in our isolates16. Other differences were reported by Serwacki et al. who note the presence of metallo-β-lactamases (including blaOXA−NDM), while they were absent in our cohort50; this further highlights the regional and temporal variation demonstrated by A. baumannii resistance in Poland. Similarly, Serwacki et al. indicate a higher prevalence of blaTEM (41%) than in our present study (24%)50, again illustrating the dynamic nature of resistance gene profiles in A. baumannii populations.

In the present study, the clinical isolates of A. baumannii collected from Polish hospitals revealed a significantly higher percentage of resistant strains than those from pig manure; this difference probably reflects the higher antibiotic pressure in the clinical settings. While environmental isolates, particularly those from food production, appear to pose a lower direct risk to hospitalised patients, their genetic profiles may still provide a valuable epidemiological context for understanding the diversity and resistance mechanisms of A. baumannii. One significant factor in such analyses involves the presence of intrinsic genes such as blaOXA−51, which was found in 98% of our A. baumannii isolates. This high prevalence confirms its role in species identification and requirement for genotypic screening beyond phenotypic assessment17. Although a key marker, the contribution of the blaOXA−51−like gene to carbapenem resistance is conditional, often requiring the presence of insertion sequences such as ISAba1; as such, its detection alone does not imply resistance17.

Notably, a limited correlation was found between STs and carbapenem resistance gene profiles. ST195, a globally-distributed MDR clone, contained strains with blaOXA−51 and blaOXA−23 (o2251, o2252, K7, K12, K16). This profile was also shared by the novel ST3285 (K18), suggesting independent acquisition of resistance mechanisms. However, the blaOXA−23 gene has frequently been identified in ST195 isolates51. Furthermore, blaOXA−40 and blaOXA−51 have been noted in ST348 (K10, K22) and ST425 (K19). The ST3285 isolate (K18) exhibited a complex repertoire of resistance genes (blaOXA−23, blaOXA−51, blaOXA−58, blaOXA−40, blaTEM), indicating that multiple resistance determinants may accumulate.

One limitation of this study is that colistin susceptibility was determined by E-testing. Although E-tests are widely used in clinical laboratories, they may give false susceptibility or resistance results due to poor diffusion and adsorption of colistin, and their use is not recommended by EUCAST due to unreliable MIC values. When possible, broth microdilution (BMD) should be used. In our study, MICRONAUT BMD was applied to four isolates, with comparable results to E-testing; however, most strains were only assessed by E-testing. It is important to note that different AST methods were used for clinical and pig manure isolates, i.e. disk diffusion and E-test for the former, and VITEK® 2 for the latter. While all tests followed the EUCAST criteria, methodological variability may introduce differences in detection sensitivity and precision, especially for some antibiotic classes. This limitation should be taken into account when comparing resistance profiles between the two isolate groups. Furthermore, while PFGE remains a valuable tool for local outbreak investigations, its resolution is limited compared to sequence-based methods. The use of an 80% similarity threshold, while widely accepted, may occasionally group genetically divergent strains or separate closely-related ones, especially when band patterns vary slightly but critically. This is a known limitation of the method and has been addressed in previous reports23,52. Finally, the number of clinical isolates differs from that of manure isolates, which translates into low statistical power.

Nevertheless, the inclusion of both clinical isolates and isolates from pig manure represents a notable advantage of our study, as it allows for a comparative analysis that enhances the epidemiological relevance of our findings. Moreover, the results have significant clinical and public health implications: the identification of XDR A. baumannii isolates and the characterization of carbapenemase-encoding genes can form the basis of infection control strategies and assist in decision-making regarding treatment and preventive measures in hospital settings.

Conclusion

The present study provides a molecular snapshot of A. baumannii isolates from Poland. Overall, A. baumannii strains demonstrated diverse clonality, with the majority of isolates belonging to the globally-distributed CC92. Our data provide an update on the current status of A. baumannii populations in Polish hospitals; they also shed light on the global pathways of the spread of A. baumannii, which can be used to monitor transmission and outbreaks. Particular monitoring should be performed on newly-identified STs. Although PFGE fingerprints do not always correspond to the same STs obtained using the MLST technique, the combination of the two methods allows more accurate tracking of bacterial transmission than either method used alone.